Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/agentcomputerai/torch/2captchanpx skills add AgentComputerAI/torch --skill 2captchagit clone --depth 1 https://github.com/AgentComputerAI/torchWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00096 | $0.01949 |
| Opus 5 | $0.00048 | $0.00975 |
| Sonnet 5 | $0.00019 | $0.00390 |
| Haiku 4.5 | $0.00010 | $0.00195 |
Grade A, and why
2captcha scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 252 lines — stays where its author put it; the contents beside it link to each section on GitHub.
2Captcha
2Captcha is a human-backed CAPTCHA solving service. You send it a captcha challenge (usually just a pageurl + sitekey), it returns a token ~20-40s later, and you inject that token into the target page to bypass the captcha.
Use this skill when:
- Stealth mode + headed Chrome fail against Cloudflare Turnstile
- reCAPTCHA v2 / v3 blocks the flow
- hCaptcha, Geetest, FunCaptcha appear on signup/login pages
- You need image-text OCR fallback
For cheaper pricing (~$0.60/1K vs $1.00/1K) and faster solves, also check capmonster — the APIs are similar enough to swap providers.
Installation
npm install @2captcha/captcha-solver
Setup
import TwoCaptcha from "@2captcha/captcha-solver";
// Default polling interval: 5000ms
const solver = new TwoCaptcha.Solver(process.env.TWOCAPTCHA_API_KEY);
// Or with custom polling (seconds)
const solver = new TwoCaptcha.Solver(process.env.TWOCAPTCHA_API_KEY, 10);
Always read the key from process.env.TWOCAPTCHA_API_KEY — never hardcode it. Torch loads .env from the repo root automatically.
Check balance
Do this first in development to confirm your key works and you have funds:
const balance = await solver.balance();
console.log(`2Captcha balance: $${balance}`);
2Captcha charges only for successful solves, but verify balance before starting large scrapes.
Finding the sitekey
The sitekey is a public identifier the target site passes to the CAPTCHA provider. Find it before calling the solver:
| Captcha type | Where the sitekey lives |
|---|---|
| reCAPTCHA v2 | <div class="g-recaptcha" data-sitekey="..."> or iframe[src*="recaptcha"] URL |
| reCAPTCHA v3 | grecaptcha.execute('SITEKEY', ...) in page JS |
| Turnstile | <div class="cf-turnstile" data-sitekey="0x4AAA..."> or iframe[src*="turnstile"] |
| hCaptcha | <div class="h-captcha" data-sitekey="..."> |
Extract it from Puppeteer:
const sitekey = await page.$eval(
'.g-recaptcha, .cf-turnstile, .h-captcha',
(el) => el.getAttribute('data-sitekey')
);
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 252 lines · 96 tokens per session scan A 1c414da89af6
2captcha is a skill published in the GitHub repository AgentComputerAI/torch (5 stars, last pushed 4mo ago), licensed MIT. It adds 96 tokens to every session and 1,949 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
br3eze-code
AgentOS uses a modular 3-tier ReAct reasoning engine. The AI (Gemini 2.0 Flash / Claude Sonnet 4.6 / GPT-4o / Ollama) automatically discovers, chains, and executes these skills via the Tool Registry and Skill Loader.
agentos-security-remediation
Review and remediate Dependabot and dependency vulnerabilities in AgentOS repositories. Use for critical or high-severity alert triage, dependency-path analysis, specialist assignment, secure upgrades, regression testing, rollback planning, release gates, and audit evidence.
agentos-specialist-swarm
Coordinate AgentOS specialist work across Planner, Engineer, Accountant, Secretary, Procurement, Expeditor, Designer, and Draftsman roles. Use for role onboarding, WBS delegation, skill selection, activity tracking, handoffs, approvals, evidence, channel rendering, procurement exceptions, and project delivery…
agentos-expeditor
Coordinate AgentOS fulfillment and delivery work with milestones, supplier and courier tracking, exception handling, recovery options, Procurement handoffs, escalation approvals, evidence, and receipt closure. Use for order delays, shipment tracking, vendor exceptions, and delivery recovery.
agentos-role-skill-factory
Create, validate, integrate, and release separate AgentOS role-skill packages with SKILL.md instructions and index.js entry points. Use when adding a specialist role, generating role metadata, wiring WBS and approval contracts, or packaging role skills for the repository.
agentos-designer
Design user, service, and system solutions in AgentOS from scoped requirements, constraints, alternatives, prototypes, reviews, acceptance criteria, and controlled publication. Use for design briefs, concepts, UX, service flows, architecture proposals, and design changes.